Competition between scrum teams

An “agile” coach/consultant was brought into to a place where I once worked. Among other things, he said he likes to set up a little competition between scrum teams, with some (undisclosed) reward for the team with the best record of completing all tasks in the sprint, over some period of time. Here’s why I think that’s a terrible idea.

Software developers have traditionally been pretty much terrible at task estimation, being notorious for projects that are late and way over budget if they complete at all.

Agile methodologies try to address this problem by encouraging us to break large tasks down into small, well-understood, predictable tasks, which are then carefully ordered and completed sequentially.

Team members are responsible for making an estimate of how long each task will take to finish. Clearly, planning improves as the team learns to make more accurate estimates. Also, to minimize work-in-progress, we want to minimize task rollover from one sprint to the next.

On the face of it, it seems like a little competition between teams would give just a little extra motivation to get better at task estimating and sprint completion.

But thinking about just a little deeper shows four important reasons why it’s a terrible idea.

1. Destroy collegiality.

Suppose we have several teams working in the same office. Maybe I am on Team A, but we all work toward the same goal, creating value for the company.

TOday: If a colleague on Team B asks me a question, of course I will help. If I need a boost from a Team C colleague, I can rely on receiving it. We’re all better off under the Golden Rule.

Tomorrow: If my B colleague asks me, I must weigh how responding will help my team’s competitive position. If it slows me down, or helps my colleague, now I have an incentive to refuse. It’s not even self-centered — all of Team A benefits when I do.

Rewarding just some equals penalizing the rest. The net effect is to divide the staff into little silos. Horizons narrow, interests diverge, collegial mutuality disappears.

Is that what management really wants?

2. Encourage undercommit to game the system.

As ever, we want to maximize Getting Things Done. That’s an absolute, and we can measure it. We also want estimating to get better. But the proposed metric, scrum completion percentage, is a ratio: stuff Done divided by some number the team decides on and announces. If you want teams to improve their ratio, reward the team with the highest ratio.

But each team can continually improve its ratio, without ever improving accuracy of estimation: simply lower the sprint commitments. This undercuts the considerable value of sprint commitments for planning, while it fails to achieve the stated goal.

Is that what management really wants?

3. Co-opt team internal metrics.

Development is hard. Estimating is hard. We need all the help, all the tools we can get (and some we can’t get). Don’t take any of them away from us.

Why doesn’t management reward teams based on Story Points completed? Because that only encourages inflation of Story Point estimates, and debases one important tool that teams have to improve estimation.

If you look at the statistics of actual vs estimated, for a good estimator the difference will have a mean of zero and a narrow variance. Some above and some below, but most are close. Sprint by sprint, the team members use their own performance metrics to learn better and better to estimate what they can get done in a sprint.

Now what happens if you penalize the underestimates (overruns) and reward the overestimates? You encourage more conservative estimates, making your estimators less accurate, which directly contradicts the goal. Along the way it also discourages the aggressive and ambitious.

Is that what management really wants?

4. Disrespect the workers and the work.

Like most responsible adults, we want to produce, to achieve. Most of us in this industry are here not just for the pay but because we find some intrinsic enjoyment and value in what we do. We don’t need gimmicks and haranguing to get us to do our jobs, any more than we need micromanagement to replace our knowledge, experience and judgment. Just give us some clarity and guidance and get out of the way.

Do you really think think that we won’t be doing our best unless you provide us with a little “fun”? So you set up little “games” that are transparently manipulative? Do you think the nature of our chosen line of work is so unengaging, even repellent, that adult professionalism isn’t motivation enough?

There is even an inappropriately moral subtext to the term “failure to meet sprint commitments”. Of course we will work as hard and smart and fast as we can, to get the the new Frobule feature ready for that Friday end-of-sprint demo for the stakeholders down the hall. But no, we are not really committing to work nights and weekends, sacrificing health and family, doing whatever it takes to complete that sprint task, just because we estimated wrong.

Leave a Reply

Your email address will not be published. Required fields are marked *